Comparative analysis of automatic text quasi-summarization algorithms

The article presents the formulation of the problem of automatic text quasi-summarization and also discusses in detail such algorithms of automatic text quasi-summarization as the Luhn’s algorithm, Latent Semantic Analysis, TextRank and LexRank. The information completeness was evaluated for a set of abstracts generated using these algorithms. The information completeness evaluation was performed using information proximity metrics: a metric based on the Jensen - Shannon distance and cosine similarity applied to vector representations of the source text and the resulting abstracts. A statistical analysis of the obtained results was carried out.